ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO

Page created by Enrique Sandoval
 
CONTINUE READING
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
ARTIFICIAL INTELLIGENCE IN
JAPAN 2020

Commissioned by the Netherlands Enterprise Agency
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
ARTIFICIAL INTELLIGENCE
    IN JAPAN 2020
    ACTORS, MARKET, OPPORTUNITES AND DIGITAL
    SOLUTIONS IN A NEWLY TRANSFORMED WORLD .

Autor     Nicole Dirksen                       5 October 2020
Co-author Sonoko Takahashi

Embassy of the Kingdom of the Netherlands
                                               P a g e 0 | 36
Innovation, Science and Technology

info@hollandinnovation.jp
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
CONTENTS
List of Figures .................................................................................................................................................................................................... 2
Summary ............................................................................................................................................................................................................. 3
Introduction ....................................................................................................................................................................................................... 4
AI Ecosystem ..................................................................................................................................................................................................... 6
         Public sector ............................................................................................................................................................................................ 7
         Research Institutes ............................................................................................................................................................................... 8
         Private sector .......................................................................................................................................................................................... 9
The Japanese AI Market ..............................................................................................................................................................................14
     AI on a global level ..................................................................................................................................................................................14
     AI market in Japan ...................................................................................................................................................................................15
     Japan’s ambitions .....................................................................................................................................................................................17
Reinventing Japan through AI .................................................................................................................................................................19
     Developments in Japan are favorable for AI ................................................................................................................................19
         Political ....................................................................................................................................................................................................19
         Economic .................................................................................................................................................................................................20
         Society ......................................................................................................................................................................................................20
         Technological ........................................................................................................................................................................................21
         Legal ..........................................................................................................................................................................................................22
     A window of opportunity for partnerships ..................................................................................................................................23
         Strenghts .................................................................................................................................................................................................24
         Weaknesses ............................................................................................................................................................................................24
         Opportunities ........................................................................................................................................................................................25
         Challenges ...............................................................................................................................................................................................25
     AI post COVID-19 .....................................................................................................................................................................................26
         COVID-19 as inhibitor of AI development ................................................................................................................................26
         Positive impact on AI developments ..........................................................................................................................................27
         AI COVID-19 solutions .......................................................................................................................................................................28
Conclusion ........................................................................................................................................................................................................29
Bibliography ....................................................................................................................................................................................................31

Cover picture form Go Tokyo

                                                                                                                                                                                           P a g e 1 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
LIST OF FIGURES

Figure 1: Actor analysis                                                             p. 6

Figure 2 Actor analysis governmental institutions                                    p. 7

Figure 3: Electronics sector                                                         p. 10

Figure 4: Automotive sector                                                          p. 12

Figure 5: Robotics sector                                                            p. 13

Figure 6: Estimated global market value AI                                           p. 14

Figure 7: Top ten Japanese companies for AI Patent families                          p. 15

Figure 8: Changes in the fields to which AI-related inventions are applied           p. 16

Figure 9: Fusion of AI and other Related Technologies                                p. 18

Figure 10: Major inhibitors and drivers of the development of AI in Japan            p. 19

Figure 11: SWOT analysis                                                             p. 24

                                                                             P a g e 2 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
SUMMARY

AI is booming, especially in Japan. This rapport will introduce the Japanese AI situation, the relevant
actors, the market, its policy ambitions, its challenges and of course opportunities for the Netherlands.
Japans AI ecosystem exists of investments from the public and private sector combined, supporting
a research environment in which AI can flourish. The Japanese government, coordinated by the
Cabinet Office, is aided by the Council for Science, Technology and Innovation and the Strategic
Council for AI Technology. The execution of their AI policies is divided over three ministries: Internal
Affairs and Communication, Economy, Trade and Industry and Education, Culture, Sports, Science
and Technology. Looking at the private sector, three big Japanese industries can be distinguished to
be very active in AI-related developments: the automotive, robotics and electronics industries.
Among these industries several different types of intersectoral and international relationships can
be found. Of these three sectors, the automotive industry spends the most on R&D.
         The AI market can be concluded to be opportune. The global AI market estimated compound
annual growth rate between 2018 to 2025 differs from 33% to 55%. By 2025, the AI global market is
projected to be worth between 156 and 360 billion euros. The Asia Pacific region is anticipated to
overtake North America’s number one spot on the global AI market by 2025. As to date, Japan has
200 to 300 AI-related companies. Japan is number one in the world as a supplier of industrial robots
and third, after China and the USA, in AI R&D. On AI patens is Toshiba Japans highest contributor,
claiming the world’s third spot, right after IBM and Microsoft.
         Japan aims to stay a prominent player in the high-tech sector with AI as one of its vital
components. Japan wants to utilize AI in its policies to address its own societal issues through its
envisioned society of the future named Society 5.0. As AI is named a core technology, it made its way
into several policy proposals, like the Japanese Moonshot program, similar to Europe’s Horizon2020,
and the cross-ministerial Strategic Innovation Promotion Program. SCAIT has developed a strategy
especially for AI developments, consisting of three tracks. The first track is productivity, with a focus
on the enhancement of creativity and innovative services. The second track is health, medical care
and welfare coming from the urgent need to respond to its rapid aging society. Finally, the third track,
mobility aims for everyone to be able to travel freely, safely and environmentally friendly.
         Japan faces several societal issues, some of which the Netherlands has shared experiences. A
main problem for Japan is its rapid aging society, of which over 40% will be elderly in 2030. This puts
pressure on the labor force and healthcare system. This pressure is also an incentive of AI enhanced
care robots and industrial robots. Of course, as this rapport is written during the global COVID-19
pandemic, the world has and still is rapidly changing from starting this project to its finalization.
Several new developments have spurred in the last months. Related to AI, the pandemic seems to
cause in some sector short-term budget cuts, but also sparked AI driven solutions and motivated
faster digitalization of traditional physical activities in which AI plays a fundamental role.
         As to conclude, despite the pandemic, AI is still booming. Japan provides several
opportunities for the Dutch private sector, for governmental institutions to exchange best policy
practices and for researcher to broaden their expertise.

                                                                                         P a g e 3 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
INTRODUCTION
Artificial Intelligence, or AI, is already entangled in our ordinary daily activities - like navigating,
recommending news articles or setting your alarm clock automatically on workdays - and will play
an important part in societies future design. The rapid technological developments and increased
interests from national and international governmental institutions, research facilities and of course
the private sector, created new opportunities as well as challenges, asking actors to respond and
adapt. Japan aims to design the society of the future, by developing a long-term strategy to adapt to
the new opportunities that high-tech developments provide. Whereas it used to be common practice
to collect information and let humans draw the conclusions from this data themselves, we are now
evolving into a society were humans and systems are connected in a virtual space. AI is a key
technology able to analyze huge amounts of data and translating it conveniently back to the human
users. AI is therefore crucial in transforming the information society to the society of the future, also
known in Japan as Society 5.0; "A human-centered society that balances economic advancement with
the resolution of social problems by a system that highly integrates cyberspace and physical space."
(Cabinet Office, 2020).

But what do we mean when we say ‘Artificial Intelligence’? Due to the multifunctionality of AI
technology, the terminology for AI is often intertwined with other concepts. The European Union
defines AI as:

                       ‘Machines that are able to learn, reason, and act for themselves.’1

Within this context, this report focuses on the current situation and therefore on Artificial Narrow
Intelligence (ANI). This refers to AI systems which are able to perform specific tasks autonomously,
but can only operate within the range of the jobs the technology is programed to do. Even the most
complicated AI technology created to date is still ANI, also known as reactive or limited memory
machines. ANI also includes AI that makes use of machine learning or deep learning to improve their
own performances, like virtual assistants or self-driving cars. ANI is not AI that can function
completely like a human being (Artificial General Intelligence) or that possesses self-awareness
(Artificial Superintelligence), of which the first is currently still a work in progress and the later a
hypothetical concept not likely to be realized in the coming decades (Forbes, 2019). The majority of
AI is currently making use of machine learning, based on statistical patterns gathered from large data
sets. AI can learn from this data to make predictions and create useful insights (Global Orange, 2020).
This ever-increasing amount of data gathered through our devices, sensors and online activities are
collected in a virtual space, also known as cyberspace. The data is analyzed by AI and the results are
fed back to us in our physical space as well. Examples are the introduction of the self-driving car able
to avoid traffic jams or the virtual assistant which can make personalized recommendations. With
this AI technology, Society 5.0 aims to meet the specific needs of each individual. The 6th Science and
Technology Basic Plan for 2021-2026, a Japanese five-year policy plan directing towards a Society
5.0 (Kuczynska, 2019), will be published this year. Therefore, this year will also mark the end of the
Japanese 5th Science and Technology Basic Plan (CSTI, 2015). Looking back at the developments

1   As used by the European Union (European Union, 2019).
                                                                                             P a g e 4 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
during the 5th plan and the set of goals for the future, provides insights in possible opportunities and
      perhaps a closer Japanese-Dutch relationship in the AI field. To provide insights and prepare Dutch
      actors interested in the Japanese AI developments, this report will highlight the relevant Japanese
      actors, market and political and economic context. By presenting what the expected developments
      as well as barriers are, this report provides keynotes for whoever is interested in exploring the
      prospect of the Japanese AI ambition.

      To help those who are newly interested to understand the wide variety of AI-related practices,
      included below is a quick overview of the most prominent techniques and applications of AI
      technologies in Japan and examples of AI applications daily used by the average consumer. These
      techniques are strongly intertwined and support each other’s functions, making the diction in some
      cases seemingly arbitrary. These six were elected as they will be mentioned often in the report:

                      At the core of AI and the basis of                          Deep learning is a more
                      many of the other examples. This                            complicated version of machine
                      technology makes it possible for a                          learning, stacking multiple layers
                      machine to learn and from past                              of algorithms between the input
                      experiences and deliver output                              and output layer. This makes it
                      based on provided data, without                             possible for deep-learning to make
                      being programmed to do so                                   intelligent decisions based on
                      repeatedly.                                                 several hierarchical concepts.
Machine Learning      Example: Google calculating travel       Deep Learning      Example: Personalized Netflix
                      time based on live traffic data                             recommendations

                      Speech processing or recognition                            Natural language processing is a
                      allows a machine to recognize and                           more sophisticated type of speech
                      interpret human speech. This                                recognition. By using deep-
                      technique makes it possible to                              learning technology, it has the
                      translate words to text, similar to a                       ability to determine the intent of
                      transcriber. It can also translate                          the spoken words. This unlocks
                      simple commands from humans to                              the possibility to have dialogues
                      actions understood and performed                            with machines hard to distinguish
                      by machines.                            Natural Language    as artificial.
Speech Recognition
                      Example: Apple’s Siri calling a            Processing       Example: Personalized auto-
                      contact by voice demand                                     complete function in WhatsApp

                      Image processing is using                                   Computer vision ‘understands’ the
                      algorithms to enhancing, restore,                           meaning of an image. This allows
                      compress or analyze an image,                               machines to process images and
                      allowing to extract more                                    recognize the information
                      information from an image.                                  presented by its visual features.
                      Example: Google translating text                        Example: Facebook detecting and
Image processing      directly with your phone’s camera       Computer vision censoring prohibited pictures

                                                                                               P a g e 5 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
AI ECOSYSTEM

In Japan the AI field is not dominated by just one type of actor, but exists of a collaboration of actors
in the public, private and research domains. This chapter will elaborate on the most prominent actors
in the AI field and how they relate to each other. The actors are divided among three groups: the
private sector, public sector and research facilities. These three groups are intertwined with each
other and the distinction between the groups made in this chapter is schematic. It must therefore be
noted that this division between three types of actors is a simplification of the actual situation. Only
the most prominent connections are shown here. It does not exclude the possibility of, for example,
a Japanese university provided with funds from the automotive sector within the AI domain, even
when it is not shown in the figure below. The reality of the wide application field of AI combined with
complex intertwined relations among the actors, demands a simplification to be comprehensible.

Figure 1: Actor analysis

Japan is especially well developed in AI-related technology in the automotive, robotic and electronic
domains, hence highlighting these industries in the figure above. This overview is a summarized view
of the relations between actors. In the rest of this chapter we will dive into the connections from the
different type of actors’ perspectives and their (individual) relations to AI technology. First, the actors
in the public sector will be described followed by the research institutions. Since there has been an
in-depth analysis for the public and private sector, the research institutions related to these have
automatically been described in the actor analysis of the public sector and, where appropriate,
mentioned in the private sector as well. Therefore, the research institutions do not have a separate
in-depth analysis with a figure of the research actors, as it would not contain additional information.
It does however have a description of the main actors and their specializations. Finally, this chapter
will conclude with a closer look to the private sector, mainly focusing on the electronic, automotive
and robotic sectors.
                                                                                           P a g e 6 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
PUBLIC SECTOR
Within the Japanese government, AI is an important method for the future design of the nation, also
known as Society 5.0. AI is part of the national agenda and financially supported in its development
(Ishii, 2018). (More on the AI roadmap of the Japanese government in the next chapter.) An example
of this is the research program PRISM, consisting of public and private R&D investments (Stronks,
2019) or the Moonshot program similar to Europe’s Horizon2020 (Kuczynska, 2019). The Japanese
government seems to give more often coordination or advise, opposed to funds, differing from the
Netherlands. The figure bellow shows the different relationships between the actors, variating from
funding /financially supporting, policy / coordination (mandatory guidance) or advise / informing
(not mandatory guidance). The most prominent actors will be presented cursive in the text.

Figure 2 Actor analysis governmental institutions

The Cabinet Office has a coordinative role regarding strategic matters under supervision of the
Cabinet Secretariats highest official, the Chief Information Officer (CIO) (Ishii, 2020). The
coordination of efforts and information exchange is the responsibility of the Council for Science,
Technology and Innovation (CSTI) (Kuczynska, 2019). Under the leadership of the Prime Minister
and the Minister of State for Science and Technology Policy, CSTI overlooks all of the nation’s science
and technology, advises and formulates policies (Cabinet Office, n.d). The execution of the AI strategy
is divided over three ministries: The Ministry of Internal Affairs and Communication (MIC), the
Ministry of Economy, Trade and Industry (METI) and the Ministry of Education, Culture, Sports,
Science and Technology (MEXT). MIC is in charge of the societal impact and governmental necessities
regarding AI. It is especially focused on supporting R&D. METI is tasked with the position of AI in the

                                                                                        P a g e 7 | 36
ARTIFICIAL INTELLIGENCE IN JAPAN 2020 - Commissioned by the Netherlands Enterprise Agency - RVO
private sector and supports related start-ups. Finally, MEXT is in charge of educational and research
aspects, promoting research as well as new study programs and other actions to challenge the human
resource shortage (European Union, 2019).
         The Strategic Council for AI Technology (SCAIT) acts as a control tower between the
ministries and manages five national research and development agencies and three research centers
(SCAIT, 2017). Finally, two independent governmental agencies relevant as well: the Japan Science
and Technology Agency (JST) and the New Energy and Industrial Technology Development
Organization (NEDO). JST targets technological developments on the governmental agenda and funds
basic research, supports the commercialization of new technologies, distributes information on
science and technology and promotes international joint research and human resources (JST, 2020).
NEDO funds technology development and coordinates technological capabilities and research
abilities of industry, academia and the government. It also promotes development of innovation and
high-risk technologies and has several international offices (NEDO, 2020).

RESEARCH INSTITUTES
AI is booming and therefore research facilities, universities
and school programs are increasingly supported with funds,
                                                                    WWII through Image Processing
staff and programs by the public and private sector. The
three most prominent research facilities are; AIRC, based at         18-year-old Niwata works together
AIST and funded by METI; the AIP Center based at RIKEN and              with the University of Tokyo to
funded by MEXT; and AIS based at NICT and funded by MIC               colorize photos using AI. With the
(Ishii, 2018). The reason for these differences in ministerial       75th anniversary of WWII’s end, this
support can be found in the focus of the centers.                        project hopes to spark happy
         AIRC is focused on AI in the private sector                memories from before the war by the
                                                                       rapidly aging generation. AI can
(Kuczynska, 2019). It concerns itself with AI in mobility,
                                                                    recognize and color natural elements
healthcare, productivity, and infrastructure. AIRC also
                                                                         and color them automatically.
collaborates with universities in Germany, France and the
                                                                    Human made objects like clothes are
UK.                                                                 still to difficult to automatically color
         AIP is specialized in generic technologies and AI in        correctly for the AI technology used.
society. It has partnerships with several EU institutes and                  (Japanese Times, 2020)
collaborates with the private sector through the coalition of
NEC, Toshiba and Fujitsu (European
Union, 2019).
         Finally, AIS looks at brain                   Student startup for AI inspections
architecture, neuro computing and
                                              Students form the National Institute of Technology work
data knowledge integration. The
                                            together with the University of Tokyo and started a venture
focus is on communication, for
                                           company for AI inspection systems targeting overhead power
example         natural      language       lines. The system analyzes footage of powerlines, detecting
processing, multilingual translation                      damage human eyes cannot see.
and multilingual speech processing                           (Shikoku Shimbun, 2020)
(Kuczynska, 2019).

                                                                                               P a g e 8 | 36
While these three centers are expected to be main links between universities, industry and
international organizations, there is also an increasing number of independent AI research facilities
with numerous collaborations. Looking at the publications, Japan is prominently represented in the
field of computer vision, robotics, speech processing and control methods. Especially AIST and NICT
can pride themselves in their developments in machine learning, computer vision, speech processing
and natural language programming. At university level, most universities offer a specialization or
study program in AI, although the University of Tokyo is the most prominent. The Japanese
government offers students and staff exchanges and scholarships. Their focus is on attracting human
resources through exchange programs like Vulcanus (European Union, 2019). Finally, private sector
has its own research centers and AI programs as well, as will be shown in the next paragraph.

PRIVATE SECTOR
The private sector has a major role in AI research and development. Prominent companies like
Toshiba, Toyota, Fujitsu, Hitachi and NEC are invested in diverse practices and technologies of AI
(European Union, 2019). The private sector is a vital part of AI R&D investments, with the automotive
sector as the biggest contributor (Greimel, 2019). The ties of the private sector with research
institutes and the government are thigh. In the AI research and development environment, the
private sector plays a vital part in stimulating innovation, research and making the AI technology
consumer friendly. It therefore plays also a major role in the Japanese developments of Society 5.0.
To elaborate more on these connections, this chapter will show an in-depth actor analysis of Japans
main industrial sectors related to AI: the electronics, automotive and robotics sectors. Of course, the
relationships and examples presented in this paragraph do not include the numerous examples of
initiatives in this rapid developing field.

Electronics sector
The first example of AI in the private sector can be found within the electronics industry through
Rakuten. Rakuten is in the Japanese electronics sector an all-round actor in internet services and
fintech. It has its own Rakuten Institute of Technology (RIT), which mainly focuses on advanced
                                                            machine learning and deep learning and
               AI disinfecting smartphones                  covers areas like IoT, network optimization,
     As electronic stores noticed an increase of infected   fraud detection, NLP, computer vision, virtual
       staff members during the COVID-19 pandemic,          reality (Rakuten, 2020). Another example of
  Telecommunications firm KDDI launched an AI service       collaboration between the private sector and
  for stores selling smartphone devices and comparable      research institutes within the electronics
        displayed items which customers are keen on         sector is the collaboration Panasonic and
    touching. Through store cameras, AI can identify the    several Japanese universities, oversees
   smartphones at display and detect which surfaces are     universities (non-EU) and public research
     touched by customers. This technique collaborates      centers like NICT (Panasonic, 2020) and AIRC
    with the robot’s sectors. After the toughed areas are
                                                            (AIRC, 2020). The nature of these
  detected on screen, a robot makes a round through the
                                                            collaborations is mainly focused on
     store to disinfect the phones using ultraviolet light.
                          (NHK , 2020)                      establishing AI research labs.

                                                                                          P a g e 9 | 36
The third example of collaboration is the interesting fusion of Fujitsu, NEC and Toshiba with the
Ministry of Education, Culture and Sports, Science and Technology (MEXT), establishing the RIKEN
Center for Advanced Intelligence Project. The companies offer funding and employees to this institute
(Stronks, 2019). Zooming in on Toshiba, Fujitsu and NEC individually, it becomes clear that their AI
affiliations do not end with RIKEN.

Figure 3: Electronics sector

         Toshiba has its own scholarships for student from Japan and several international
universities from Europe (although none from the Netherlands) through the Toshiba International
Foundation (Toshiba, 2020). Furthermore, there has been speculation of Toshiba opening its own
scholarship specifically aimed at AI researchers with the University of Tokyo, although this has not
been confirmed (Japan Today, 2019).
         Fujitsu is specialized in big data. Its AI endeavors are bundled together in Zinrai, Fujitsu’s
Human Centric AI center with a focus on detection, recognition and machine learning. Fujitsu’s
ambitions spread outside the electronics sector, as it also wants to use AI in the robotics and
automotive sector (Ishii, 2018). Fujitsu has a global partnership with Microsoft to accomplish this
(Fujitsu, 2020).
         Finally, NEC has put its AI endeavors in their IoT platform named NEC the WISE. This platform
is aimed at businesses and AI makes it possible to collect and process data efficiently at high speed
to create an online infrastructure for the businesses (NEC, 2020). Apart from this platform aimed at
businesses, NEC has also developed AI applications for governmental institutions. Examples being a
police fingerprint system and a facial recognition system for the Japanese immigration services (Ishii,
2018).

                                                                                       P a g e 10 | 36
It becomes apparent that the electronic sector is intertwined with actors outside its own electronic
bubble, collaborating with European universities and international players like Microsoft. This sector
is highly interconnected among the AI application field, creating more than technology aimed as
consumer products. The electronic sector invests in long term knowledge development and
facilitates the governmental agenda, but is not unique in this approach. This is to a certain extent
similar to the automotive sector.

Automotive sector
The actors within the automotive sector have a strong interconnectedness among each other and
international endeavors similar to the electronics
sector. An example is the strategic partnership of                        Toyota Woven City
Alliance Venture, a collaboration of Mitsubishi Motors,
                                                               “Imagine, a fully controlled site that would
Renault and Nissan, with its headquarter in Amsterdam.
                                                              allow researchers, engineers and scientists
It pursues strategic investments at all maturity stages in      the opportunity to freely test technology,
startups developing disruptive technologies or                such as autonomy, mobility, robotics, smart
businesses (Alliance Ventures, 2020).                          home connected technology, AI, and more,
         However, the most prominent automotive                   in a real-world environment. (…) We
actors in this story are Toyota, Honda and Nissan. To         thought, why not build a real city, and have
put the role of the automotive sector in perspective, this       real people live in it, and safely test all
sector has occupied the top four slots for of overall R&D       kinds of technology? (…) This would be a
spending in Japan in 2019. Toyota being the biggest               truly unique opportunity to create an
                                                                  entire community, or “city,” from the
spender with a budget of €8.6 billion and the Toyota-
                                                                    ground up and allow us to build an
affiliated Denso as the fourth highest with €4.1 billion.
                                                                    infrastructure of the future that is
Toyota was followed by Honda with €6.7 billion and               connected (…) It would be a chance to
Nissan with €4,3 billion (Greimel, 2019). For AI specific,      collaborate with other business partners
the investments of Toyota, Nissan and Honda are mostly          and to invite all interested scientists and
directed at the development of self-driving cars.                  researchers from around the world.”
         For Toyota the necessary AI technology for self-                     ~ Akio Toyoda,
driving cars, deep learning of facial, behavioral and           President   of  Toyota Motor Cooperation
speech recognition, is being developed in the Toyota
Research Institute (TRI). Honda cooperates with the                            (Toyota, 2020)
Chinese SenseTime company that specializes in deep
learning and image recognition. Honda is not only active in the automotive sector, but also in the
robotics sector as it also uses its AI endeavors on general physical mobility (Honda, 2020). Finally,
researchers from Nissan cooperate with NASA to develop AI technology, sensors, and software to
build a self-driving robot car (Ishii, 2018). Nissan uses robotic technology from NASA to improve its
product and, like to Honda, is also intertwined with the robotics sector, although in this case the
robotic technology is not from Japanese origin.

                                                                                           P a g e 11 | 36
Figure 4: Automotive sector

Robotics sector
The actors within the Japanese robotics sector seems to be more nationally focused compared to the
previous sectors, but have big interdisciplinary players. Our first example of AI in the robotics sector
is Mitsubishi Electric’s Maisart research program. This program focusses on developing and
researching deep-learning, reinforcement learning and big data analysis (Mitsubishi Electric, 2020).
A practical application would be in industrial robots, where AI technology can distinguish
abnormalities and detect signs of machinery failure prior to actual breakdowns (Mitsubishi Electric,
2019).
        Secondly, Hitachi is, similar to Mitsubishi Electric, a broadly oriented company with ties to
the automotive, electronics and robotics sector. Hitachi sees robotics as the bridge between the
virtual world and the consumers. It uses AI to create robots that can navigate, communicate and
interact with its surroundings (Hitachi, 2020). Another actor in the robotics sector is FANUC. FANUC
specializes in industrial robots and factory automation. It uses AI machine learning to train robots in
an easier manner. Hitachi and FANUC joined hands with Preferred Networks to establish a joint
venture: Intelligent Edge Systems (IES). IES utilizes AI as an intermediary between the cloud and
robotics to achieve cyclic real-time control. IES has stated to be committed to the realization of a
human-centered Society 5.0 (Preferred Networks, 2018). On the topic of robotics AI partnerships,
another partnership can be found between Yaskawa Electric Corporation and XCompass in the form
of a new company named AI Cube. AI Cube develops AI technology for manufacturing and industrial
robots as well (Yaskawa, 2018).

                                                                                       P a g e 12 | 36
Figure 5: Robotics sector

                                         A final example of AI robotics collaboration can be found in
                                         OMRON’s creation of the table tennis robot Forpheus,
             Stacking Robots             combining robotics with AI and the gaming industry. OMRON
                                         is specialized in industrial automation and electronic
    In August 2020, a robot was tested
     stacking sandwiches, drinks and
                                         components. For this project OMRON joined hands with
        meals on the selves of a local   Square Enix, a famous Japanese gaming company with well-
    convenience store. In this stadium   known titles like Final Fantasy (OMRON, 2020). The goal of
      the robot is remotely controlled   their combined research is to develop an AI algorithm that
   through VR goggles, but aims to use   generates motivational feedback to bring out the ability of
  AI to mimic human movements in the     human growth, also known as serious gaming. OMRON
    near future. Depending on the test,  provides AI technology that can read human emotions.
  Lawson is planning to deploy its first Square Enix AI technology can analyze (emotional) responses
     robot in September this year and
                                         from users and predict the game’s unfolding. Both have
     FamilyMart plans to use stacking
                                         different motivations for this project. Square Enix wants to
              robots in 2022.
          (Japanese Times, 2020)
                                         create a way of playing in the in the real world without the
                                         need of a display screen and OMRON wants to develop future
                                         factory automation, healthcare and social solutions. However,
both believe in the establishment of a new relationship between humans and machines, in which the
machine can optimize human performance. With this shared vision of AI being used for optimizing
human performances, well fitted in the Japanese AI strategy which will be explained in the next
chapter, this actor environment overview is concluded. While the next chapter will not go in-depth
into individual actors, it will become clear why we see an increase in start-ups and (cross sectoral)
collaborations in Japan.

                                                                                     P a g e 13 | 36
THE JAPANESE AI MARKET

AI-driven technologies are the next step in the technological revolution, some would even describe it
as the next disruption in enterprise technology (OSA DC, 2018). But what does this mean for the
market? This chapter will describe AI in a global perspective and its forecasts for Japan. This chapter
contains multiple sources of market analysis of which all project growth of the future AI market.
However, it must be noted that this report is being written during the COVID-19 pandemic. Meaning
that the projections up until now have been estimations based on data gathered before COVID-19’s
impact on the world’s economy. This chapter will therefore also discuss the potential of AI in a post-
COVID-19’s world. Finally, a PESTL and SWOT analysis will be added to gain a complete overview of
the inhibitors, drivers, strength, weakness, opportunity and threat of AI in Japan.

AI ON A GLOBAL LEVEL

While the exact projections of AI revenue differ among sources, all agree: AI will be booming. The
global AI market estimated compound annual growth rate (CAGR) between 2018 to 2025 differs
from 33% to 55%. By 2025 the AI global market is projected to be worth between 156 and 360
billion euro’s2. Although it must be noted that the projections do not make statements directing
towards this increase of market value to be linear.

        In billions

Figure 6: Estimated global market value AI

2   Based on projections from AMR, GVR and Fortune Business Insights. Dataset included in figure.
                                                                                             P a g e 14 | 36
On a global scale North America has held the dominant share in
                                                                        AI helps business recovery
the AI market. As one of the early adopters, North America
adopted the AI developments in a faster pace than most regions.        The Japanese Odajima invented
However, the Asia Pacific region is anticipated to overtake North            an app that predicts a
America’s number one spot on the global AI market by 2025 (GVR,        restaurant’s customers amount
2020). Several sources have estimated the Asia Pacific region to       through AI with 95% accuracy.
have the highest compound annual growth rate in the AI field,               Odajima says AI-based
projecting a CAGR up to 59,4% from 2019 to 2025 (AMR, 2020)in          management    can raise sales per
                                                                       customer significantly and help
the most aggressive case.
                                                                        restaurant holders to calculate
        In addition to the increased market share of AI, the
                                                                          business survival methods
different AI technology fields have increase and changed in                  during the pandemic
importance as well. AI software in 2020 is estimated to grow                  (Tsukimori, 2020).
154% annually with a market value of 21 billion EUR (Shanghong,
2020). As of the past years, natural language processing has been the number one adopted AI
technology on a global scale (Fortune Business Insights, 2020). However, it is projected that machine
learning will gain a more prominent role by 2025 (GVR, 2020). As part of the machine learning
technology, deep-learning is expected to increase from a 186 million EUR in 2015 to a 10.2 billion
EUR in 2024 (OSA DC, 2018) An increase in natural language processing, image processing and
speech recognition is expected as well, although not with such a rapid increase of revenue as machine
learning. Another projection is the rise in demand for AI in healthcare, automotive and AI powered
industrial robots. AI revenues in the healthcare sector and automotive industry, mainly in self-
driving cars, can be expected to increase with several billions (GVR, 2020) (OSA DC, 2018).Industrial
robotics, a source of pride, gets an ever more prominent place, as Japan´s manufacturers deliver 52
percent of the global supply (IFR, 2020).

AI MARKET IN JAPAN

Japan has at the moment 200 to 300 AI-related companies (Data Artist, 2020). Japan is number one in
the world as a supplier of industrial robots and third, after China and the USA, in AI R&D (OSA DC,
2018). Zooming in on Japanese companies AI patents, Toshiba is Japans champion and the world’s
third, after IBM and Microsoft.

                                                   2683                               Toshiba
                                                              5223
                                            2772                                      NEC
                                                                                      Fujitsu
                                     2890                                             Hitachi
                                                                         4406
                                                                                      Panasonic

                                   3487                                               Canon
                                                                                      Sony
                                                                        4303
                                                                                      Toyota
                                        3959                                          NTT
                                                                4233                  Mitsubishi
                                                   4228

 Figure 7: Top ten Japanese companies for AI Patent families (European Union, 2019)
                                                                                                P a g e 15 | 36
Research results from METI show a rapid rise in domestic patent applications for AI-related
inventions as well. Their latest numbers show a 54% increase in 2018 compared to 2017, identifying
it as the ‘third AI boom’. In most of the inventions machine learning is used as the main technology,
referring increasingly to deep-learning over the past years. Figure 8 shows the most patented AI
applications in Japan (METI, 2020).

Figure 8: From METI report: Changes in the fields to which AI-related inventions are applied (METI, 2020)

In the future, Japan`s position within the AI market is estimated to grow up to 0,7 to 0,75 billion EUR
by 2030 (Graci, 2020) (OSA DC, 2018) and expected to gain an economic return of 1,1 billion EUR by
2045 by the Japanese government. As will be described in the next paragraph, one of the focus areas
for Japan is mobility, which is expected to increase a lot. This is reflected by the growth in the transport
sector, expected to increase the most with 0,26 billion EUR and the growth of 0,10 billion EUR in the
manufacturing sector, which also includes self-driving cars (OSA DC, 2018).

                                                                                                       P a g e 16 | 36
JAPAN’S AMBITIONS

Japan aims to stay a prominent player in the high-tech sector with AI as one of the vital components,
not just for the sake of being a leading nation in this world changing technology. Japan wants to utilize
AI in its policies to address its own societal issues, to have sovereignty over its fate and to actively
design its own future. Therefore, the Japanese AI strategy is a key part of the transition to their
envisioned Society 5.0 (Cabinet Office, 2020). Society 5.0 stands at the core of several research
programs, like the Japanese Moonshot program, similar to Europe’s Horizon2020, and the cross-
ministerial Strategic Innovation Promotion Program (SIP) (Cabinet Office, 2020) (Cabinet Office,
2019). While AI plays a supportive role in many developments within the Society 5.0 vision and
related programs, SCAIT has also developed a strategy especially for AI developments and a roadmap
as how and where to implement its different applications. This paragraph will explain the AI strategy
and introduce Japan`s main ambitions.

So how is the AI development visualized? The Japanese government has estimated several priority
areas 3 as part of the roadmap for AI integration in society and industry (see figure 7).
         The first area is productivity, with a focus on the enhancement of creativity and innovative
services. User-driven hyper customization by automation and enhancement of production and
services, distributes goods and services consistent with the needs of citizens. One clear future end
goal is the utilization of robots that are autonomously able to predict and produce ‘just in time’ and
‘on demand’, resulting in a zero-waste society. The second priority area is health, medical care and
welfare. Japan aims to be the leader in medical care and welfare technologies by utilizing big data and
AI. This goal comes from the urgent need to respond to its rapid aging society. This part of the strategy
is very active in prevention and monitoring of diseases. In the future, Japan wants to provide the
possibility to design our own body and to replace body functions by artificial organs and sensors.
Further, robots will function like family members, being part of our daily lives, caring for us and our
day to day tasks. Finally, mobility is the third focus point, with the aim for everyone and anything to
be able to travel freely, safely and environmentally friendly. From decreasing the human factor in
accidents by autonomous transportation or minimizing the need of transportation altogether,
mobility will change with AI. SCAIT foresees a future where the cyber and physical space are fused,
with examples as virtual tourism and virtual office spaces (SCAIT, 2017).
While these ambitions seem quite futuristic, this aforementioned strategy and the examples are not.
SCAIT introduced a roadmap with three phases how AI technology is expected to integrate in society.
The first phase is the growth of utilization and application of data-driven AI along the needs of the
service industries. We are currently gradually transitioning from the first phase into the second
phase. The phases do not have the same starting point for all AI-related industries, and the transition
takes place with different paces. The second phase is about the public use of AI and data. The roadmap
considers some other, unanticipated use to become relevant as well. The third phase, is expected to
arrive between 2025 -2030. In this phase, multiple domains get connected and transition into a cross-
domain ecosystem.

3 These priority areas are part of the Industrialization Roadmap. In addition to the three areas, a fourth area
“information security” was mentioned as a cross-sectional area, but not established as a roadmap target.
                                                                                               P a g e 17 | 36
Figure 9: Fusion of AI and other Related Technologies (SCAIT, 2017, p. 4)

The AI strategy and roadmap has been written taken changes in technology and societal preferences
into account. However, what could not have been predicted was the devasting effect COVID-19. The
fallout of the virus has impacted our lives broadly and the timing of the neatly planned phases of the
AI implementation. Furthermore, the implementation of AI is also more than a market analysis of
intellectual property and economic returns. AI has a potential impact on many aspects of society. The
next chapter will therefore provide an in-depth look in the opportunities and threats for AI and dive
into the macro-level trends of the Japanese AI market from several perspective, which includes the
(anticipated) effects of the COVID-19 virus as well.

                                                                                      P a g e 18 | 36
REINVENTING JAPAN THROUGH AI

AI has a potential impact on many aspects of society. The impact however, has not always been for
the worst. Technological developments can thrive during chaotic times and this crisis is no different.
AI developments have not come to a complete standstill until economy recovers. This chapter looks
at macro level trends, which have an effect on the development of AI in Japan by using five different
perspective; the political, economical, societal, technological and legal dimensions, or in short PESTL.
Looking through these five glasses, this rapport describes the present Japanese situation and also
includes the current uncertainties surrounding COVID-19.

DEVELOPMENTS IN JAPAN ARE FAVORABLE FOR AI

     Political                    Economic              Society             Technological        Legal

•Self-sufficiency             •Economy             •Medicare                •Security       •Data storage
                                                    demand
•Society 5.0                  • Automation                                  •Software       •Legal
                                                   •Robot friendly                           framework
                              •Talent shortage                              •Robotization
                                                   •Consumer
                                                    comfort
 COVID-19 positive effect on AI development
 COVID-19 negetive effect on AI development        •Remote services
 COVID-19 no effect on AI development

Figure 10: Major inhibitors and drivers of the development of AI in Japan

POLITICAL

     •     Self-sufficiency: Due to the geographical location and the strained relationships with its
           neighbors, Japan aims to be self-sufficient in many dimensions. This includes technology, as
           having the latest technological developments through own inventions means having more
           insight in the specifics of the technology. This is reflected in policies to become a frontrunner
           in several AI domains and investments in the many recently established AI knowledge
           centers. As the pandemic has shown that international supply-chains can be frail, it is
           expected that this desire for self-sufficiency of the island nation will increase even more.
     •     Society 5.0: AI is of strategic importance for the Japanese government as a core technology
           of the vision of Society 5.0, and with that, the 5th Science and Technology Basic Plan and the
           Annual Growth Strategy (European Union, 2019). Developing cross sector services and
           policies able to adequately accommodate Japans future, Japan wants to lead in practical
           applications of AI. This vision for the future influences the direction of AI related policies and
           supports investments in AI related project. This future vision heavily relying on digitalization,
           of which AI, as one of the key digital technologies, can be seen as the backbone, and is
           expected to become increasingly relevant due to the pandemic. Moreover, the Japanese
           government has decided to allocate finances to support R&D, including AI, even more during
           these difficult times.

                                                                                             P a g e 19 | 36
ECONOMIC

  •   Economy: At the moment Japan experiences
      an economic downturn. Many sectors are
                                                                     Tourism meets AI
      suffering due to the pandemic and have
      generally fewer financial means to invest.           COVID-19 put a halt to the Tokyo 2020
      Although digital services were less impacted     Olympic Games, international tourism and it
      compared to tourist sector or the automotive     lowered domestic tourism dramatically. The
      industry. This economic crisis creates             Japanese company Ebilab has developed a
                                                        service for the city of Ise, providing a virtual
      incentives to continue AI developments, but
                                                         reality tour which takes tourists on virtual
      that unfortunately does not help on the short-
                                                          excursions to the Ise shrine. To make the
      term.                                              virtual holiday complete, you can virtually
  •   Automation: The Japanese workforce is               shop at Ise’s souvenir store and interact
      shrinking due to an aging population and a low           with employees by web camera.
      birthrate (Gloture, 2019). Manual labor is                      (Tsukimori, 2020)
      gradually being replaced by robotization and
      automatization.     Robots     will    become
      increasingly capable of doing complex and physically demanding tasks due to advancement
      of AI, which is making the aging less problematic. During the pandemic this robotization of
      tasks has gotten an extra incentive due to safety constrains, as social distance has to be
      maintained.
  •   Talent shortage: METI foresees a shortage of 120.000 AI business experts by 2030. To tackle
      this problem, the government aims to attract 250.000 AI experts from abroad by
      internationalization of education programs and by making foreign exchanges more attractive
      with funds and special working visas (Gloture, 2019). English as a working language is
      stimulated and appointing foreign researchers in management positions is encouraged as
      well (European Union, 2019). Japan is also in the process changing their education system
      (OECD, 2018). Japan is not the only nation facing an AI talent shortage, so is the Netherlands
      among many others (Ministerie van Economische Zaken en Klimaat, 2019). This problem will
      be hard to solve soon by attracting foreign expert alone, while the effects of educational
      reforms can only be harvest in the future, if anything, COVID-19 made the immigration
      process more difficult.

SOCIETY

  •   Medicare: Japan’s elderly population will exceed the 40% in 2030. Because of the predicted
      increasing demand of medical support, Japan has set a goal to become the leader in medical
      care, health, welfare and longevity by combining medical advancements. During the COVID-

                                                                                       P a g e 20 | 36
19 pandemic the value of AI in several medical fields became clear
                                              once again; from analyzing huge amount of data to determine a
                                              feasible vaccine to remote care. This trend was already in place
                                              before the pandemic, but is expected to have been boosted due to
                                              the situation even more. This increased need of AI backed medical
                                              equipment is supported in several programs and packages from
                                              the Japanese government, paving the way for the needed digital
                                              infrastructure. One example of this is the SIP ‘AI Hospital System’,
                                              building a high-security database system (Cabinet Office, 2019).
                                              • Robot friendly: The adaptation of AI in daily routines is
                                              relatively well accepted in Japan. Whereas in Europe there are
  The aforementioned stacking robot of
                                              some sceptic and sometimes frightening visions of a future
 convenient stores resembles, according
                                              interwoven with AI, the consumer market of Japan is highly
  to its makers, a kangaroo. The earlier
model was a humanoid model, but made
                                              enthusiastic. Especially in regards to AI driven robots that can
the customers uneasy. This model makes        resemble human behavior. These are often presented as happily
       customers feel more at ease.           co-existing with humankind by popular media, portraying cute (or
                                              kawaii) robots like Doraemon rather than the Terminator. This
                 aids to the societal acceptation of robots and AI development in the services (Gloture, 2019).
                                         CNN 2020
                 COVID-19 is not expected to have significant influence on this trend.
            • Consumer demands: While a lot of AI applications discussed are targeted to solve societal
                 problems, the AI technology developed can also be used to make the lives of the consumers
                 easier. For example, AI technology originally used for virtually office meetings and translation
                 services, have now been adopted by the average consumers, especially during the pandemic.
                 As Japanese citizens seem to take less (public) issue with the collection of personal
                 information to be analyzed and used in the machine learning process, AI driven services can
                 quicker develop and adapt to the consumers need (Gloture, 2019).
            • Remote and online services: Although some of the previous concepts have touched this
                 subject somewhat already, it must be noted that the social distancing society we live in today
                 has impacted services. Services like online shopping and online meetings are backed by AI
                 technology and increasingly used during the pandemic. The bulk of consumer data produced
                 by this is only aiding the AI technological developments. Be it in the production, medical or
                 mobility roadmap, AI is expected to become more and more integrated in services. Many
                 services traditionally physically performed, will become remote controlled through AI or
                 done by AI robots, as is planned in the AI strategy (SCAIT, 2017). To put it into perspective;
                 the potential impact of AI is estimated to replace half of the labor force in Japan by AI, remote
                 controlled services and robots in the next 10 to 20 years (Lundin & Eriksson, 2016).

       TECHNOLOGICAL

           •   Security: Due to the digital revolution, relatively new security threats have become reality.
               With AI technology, anomalies in access or usage patterns can be detected, but also
               irregularities in data patterns, which for example can be used to audit accounting. This trend
               seems to be mostly determined by the universal digitalization movement.

                                                                                                 P a g e 21 | 36
•   Software: The earlier described trend of automatization as spurred software developments
        and AI. Modern products are increasingly relaying more on software updates than hardware
        replacements. That is also the case in Japan that tends to lean more towards hardware
        development than software.
    •   Robotization: Japans strategy towards robotization capitalizes its advanced industrial and
        functional robots’ skills to create ‘the most advanced robotics society’. This should also result
        in Japan as an innovation hub (Kuczynska, 2019). Robots are expected to become increasingly
        part of our day to day activities. During the COVID-19 pandemic it can indeed be noted that
        Japan is showing more examples of robotic solutions compared to a more app-oriented
        approach as we do in Europe.

LEGAL

    •   Data storage: The Japanese government is in the process of
        facilitating data sharing platforms and is creating a legal                Ending the Hanko era
        framework to accommodate this desire. Japan introduced a set           The Japanese Hanko is a small
        of guidelines regulating the circulation of data use in public and       circular red seal similar to a
        private domains (European Union, 2019). One example that              signature   and needed on official
                                                                                forms. Because of the need to
        relates to the health, medical care and welfare, is the ‘AI               manually certifying official
        Hospital System’ program under the SIP. This program aims to            applications with a company
        develop a high-security database system, supporting multiple              seal, not all employees can
        languages and ensuring data confidentiality, aiming to have           telework.   During the pandemic
                                                                               legal restrictions on electronic
        this data widely available and improve medical practice                    signatures became more
        (Cabinet Office, 2019). AI’s intelligence improves with the                         relaxed.
        amount of data. This example paves the way for AI assisted                      (Nippon,  2020)
        auto recording, documentation and AI-assisted diagnosis.
    •   Legal framework: Like most countries, the Japanese legal framework is in need of an update
        to take the newest digital developments into account. This means that it is uncertain which
        developments are allowed in future. For example, AI uses big amounts of data to optimize its
        performance. If there are less limitations for data sharing, AI has the chance to develop faster
        compared to a situation that has a limited permission to use, for example, consumers data
        (OSA DC, 2018). Japan seems to have a similar approach regarding legal and ethical matters
        as the European Union regarding AI. The EU and Japan signed a joint statement to promote a
        human-centric approach4 to AI (European Council, 2019), Japan is a loyal supporter of the
        Data Free Flow with Trust (Okano-Heijmans, 2020) and Japan finalized the Social Principles
        of Human-Centric AI in 2019 (European Union, 2019). COVID-19 increases the necessity of
        revisiting the laws in certain sectors. For example, official documents which normally needed
        to be signed in person are increasingly allowed to be digital, decreasing the amount of time
        and papers spent.

4 It must be noted that, though similar to the EU principles, the EU is more individual focus whereas Japan
looks for an overarching societal vision
                                                                                              P a g e 22 | 36
You can also read